How to get performance matrix

[Rohan Kangralkar]

Hi,

Thanks very much to Wendell, Gedare and Joel for the help.

The user extensions and the test examples are very usefull for gathering
the performance results.

We plan to analyze the suggested user extensions and test example and come
up with a test design. We will post test design for review as soon as we
are done with the design.

One again thankyou.


Regards
Rohan Kangralkar


On Thu, Mar 8, 2012 at 11:43 AM, Joel Sherrill <joel.sherrill at oarcorp.com>wrote:

> On 03/08/2012 08:21 AM, Gedare Bloom wrote:
>
>> On Wed, Mar 7, 2012 at 9:37 PM, Gedare Bloom<gedare at rtems.org>  wrote:
>>
>>> Hi,
>>>
>>> If your target has a decent on-chip timer then you should be able to
>>> use the tmtests to get some idea. I can't speak to how
>>> accurate/precise the numbers would be but it would be a good place to
>>> start.
>>>
>>
>
> The numbers from the tmtests are as good as the timer driver
> and underlying hardware. :)
>
> On some boards, they report cycles, on others microseconds.
>
> Various tmtests and psxtmtests include interrupt and context
> switch timing.  These are straightforward tests generally designed
> to reliably give accurate timings for a specific use case of a method.
> The focus is on determining if a change has an impact and being
> able to compare RTEMS performance across BSPs and architectures.
>
> If you have other timing scenarios you are interested in, then
> we are open to submissions. This type of code needs careful
> review to ensure you are accurately measuring what you think
> you are and don't end up with anomalies.
>
> --joel
>
>  On Wed, Mar 7, 2012 at 7:24 PM, Rohan Kangralkar
>>> <rohankangralkar at yahoo.com>  wrote:
>>>
>>>> Hi All,
>>>>
>>>> RTEMS is a very deterministic RTOS and we would like to get the
>>>> performance
>>>> matrix for ADSP-BF527 processor (TLL6527m).
>>>>
>>>> We plan to get measurement of
>>>> 1. context switch latency
>>>>
>>> tm26
>>>
>>>  2. interrupt latency
>>>>
>>> tm27
>>>
>>>  3. footprint size
>>>>
>>> should be a binutils 'size' program for your target that you can run
>>> on the binary executable to get its size... I'm not sure exactly what
>>> you are looking for / meaning by footprint here.
>>>
>>>  Check sptests/spsize -- might be what you are looking for
>>
>>  I am looking into the details of the code to find the points of context
>>>> switch. Can someone help me in identifying the points of context switch.
>>>> Does RTEMS provides hooks at these points?
>>>>
>>>>  Good luck!
>>>
>>>>
>>>> ___________________________
>>>> Thanks and regards
>>>> Rohan Kangralkar
>>>>
>>>>
>>>>
>>>>
>>>> ______________________________**_________________
>>>> rtems-users mailing list
>>>> rtems-users at rtems.org
>>>> http://www.rtems.org/mailman/**listinfo/rtems-users<http://www.rtems.org/mailman/listinfo/rtems-users>
>>>>
>>>>  ______________________________**_________________
>> rtems-users mailing list
>> rtems-users at rtems.org
>> http://www.rtems.org/mailman/**listinfo/rtems-users<http://www.rtems.org/mailman/listinfo/rtems-users>
>>
>
>
> --
> Joel Sherrill, Ph.D.             Director of Research&   Development
> joel.sherrill at OARcorp.com        On-Line Applications Research
> Ask me about RTEMS: a free RTOS  Huntsville AL 35805
>    Support Available             (256) 722-9985
>
>
>
> ______________________________**_________________
> rtems-users mailing list
> rtems-users at rtems.org
> http://www.rtems.org/mailman/**listinfo/rtems-users<http://www.rtems.org/mailman/listinfo/rtems-users>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.rtems.org/pipermail/users/attachments/20120308/1c078e65/attachment-0001.html>